Predictive modeling of T-cell expansion: leveraging morphological metrics with invitrogen attune cytpix flow cytometer for enhanced cell therapy manufacturing 4539

Abstract Description Objective This project aims to enhance the efficiency and predictability of cell therapy workflows. By monitoring T-cell morphological metrics of unmodified T-cells across various donors during ex vivo expansion, we built a predictive model using images from the Invitrogen Attune CytPix Flow Cytometer. Method Quantitative biomarkers and morphological data from unmodified T-cell expansion across 3 fresh and 4 frozen peripheral blood mononuclear donor leukopacks were tracked. Data points on cell growth, viability, marker expression and morphology were collected on Day 0, Day 2, and Day 9/12, mimicking the critical timepoints in the CAR-T cell manufacturing workflow. To identify the cell subpopulations, we used cell graphs and community detection algorithms using the Phenograph method. Biomarker enrichment analysis was done to annotate cell subpopulations, aligned with T-cell expansion fold change. Results The fresh donor datasets used a morphology-only based unsupervised clustering method. Donors that expanded the best showed the most uniform morphology profiles on Day 0, while the donors that expanded worse showed highly diverse cell morphology profiles with parameters related to circularity, size and texture of the cells. Conclusion These insights allow researchers to screen starting donor material and provide them with tools to help make informed decisions about the manufacturing process before it begins. Funding Sources ThermoFisher Scientific Topic Categories Computational and Systems Immunology (COMP)